[Gc] GC synchronized with foreign or external threads?

Boehm, Hans hans_boehm@hp.com
Tue, 18 Nov 2003 15:00:08 -0800


A newer version of wrap.h is used by more recent versions of qprof, which
now have a real web site: http://www.hpl.hp.com/research/linux/qprof/ .

I would take the one out of version 0.4.  Testing is still spotty, but any
bugs are probably fairly obvious cut-and-paste errors, and should be pretty
easy to fix.  (Qprof uses it more extensively now, but not all with all
possible numbers of arguments.)

Hans 

> -----Original Message-----
> From: gc-admin@napali.hpl.hp.com [mailto:gc-admin@napali.hpl.hp.com]On
> Behalf Of Alec Orr
> Sent: Tuesday, November 18, 2003 2:24 PM
> To: gc@napali.hpl.hp.com
> Subject: [Gc] GC synchronized with foreign or external threads?
> 
> 
> Folks:
> 
> RH Linux 8.0, Boehm GC6.2
> 
> Last July there was some discussion about allowing the GC to 
> attach to 
> threads from foreign pre-compiled modules (.so's).  I'm interested in 
> having foreign compiled .so's (which have pthread_create 
> calls inside), 
> be suspended while the GC runs in our application.  These foreign 
> threads will also call our API's.  These API's create 
> structures which 
> reside in the GC heap.  I'm really not interested in redirecting any 
> existing malloc calls in those threads, only in suspending 
> their threads 
> while the GC runs.  I have an API that foreign threads must 
> call before 
> creating the thread, at which point I could inject 'simulation' of a 
> _REENTRANT and GC_LINUX_THREADS).
> 
> Is the wrap.h implementation suggested last August still the 
> recommended 
> path 
> (http://www.hpl.hp.com/hosted/linux/mail-archives/gc/2003-July
/000138.html)?

Best regards and thanks for any replies,
Alec Orr





_______________________________________________
Gc mailing list
Gc@linux.hpl.hp.com
http://linux.hpl.hp.com/cgi-bin/mailman/listinfo/gc